factor analysis and scale revision_reisewallercomrey_2000

Psychological Assessment 2000, Vol. 12, No. 3, 287-297 Copyright 2000 by the American Psychological Association, Inc. 1040-3590 /00/S5.00 DOI: 10.1037//1040-3590.12.3.28 7 Factor Analysis a nd Scale Revision Steven P. Reise University of California, Lo s Angeles Niels G. Waller Vanderbilt University Andrew L. Comrey University of California, Lo s Angeles This article reviews methodological issues that arise in the application of exploratory factor analysis (EFA) to scale revision and refinement. The authors begin by discussing how the appropriate use of EFA in scale revision is influenced by both the hierarchical nature of psychological constructs and the motivations underlying the revision. Then they specifically address (a) important issues that arise prior to data collection (e.g., selecting an appropriate sample), (b) technical aspects of factor analysis (e.g., determining the number of factors to retain), and (c) procedures used to evaluate the outcome of the scale revision (e.g., determining whether the new measure functions equivalently fo r different populations). Personality measurement by self-report questionnaire is a thriv- ing enterprise of critical importance to theory development and testing in many psychological disciplines such as clinical psychology. At least three journals focus on statistical analyses of questionnaire data: Psychological Assessment, Journal of Personality Assessment, an d Assessment. Many of the articles in these jou rnals us e exploratory factor analysis (EFA) and, oftentimes, the factor analytic findings ar e used to guide scale revision. In this article, we review methodological issues that arise in the application of EFA to the scale revision an d refinement process. This article begins with an overview of several issues pertinent to the application of EFA in scale revisions. Two of the topics we address are the hierarchical nature of psychological constructs and th e motivations fo r revising a scale. Methodological issues that in the of EFA to ar e covered. Specifically, we address issues that arise prior to data collection (e.g., selecting an appropriate sample), technical aspects of factor analysis (e.g., determining the number of factors to retain), an d procedures used to evaluate the outcome of the scale revision (e.g., determining whether the new measure functions equivalently fo r different populations). W e conclude by highlight- ing two additional topics: dimensionality and scale score interpretation, an d principal components versus factor analysis. Inevitably, an y review will omit important topics. This review is no exception, and to clarify ou r points of deemphasis we note the following. First, ou r review is primarily concerned with EFA, an d we do not provide a thorough consideration of alternative multi- variate models that may be helpful during scale revision, such as Steven P. Reise and Andrew L. Comrey, Department of Psychology, University of California, L os Angeles; Niels G. Waller, Department of Psychology, Vanderbilt University. Correspondence concerning this article should be addressed to Steven P. Reise, Franz Hall, Department of Psychology, University of California, Los Angeles, California 90095. Electronic mail may be sent to [email protected]. principal-components analysis (see Velicer & Jackson, 1990) or multidimensional scaling (Davison, 1994). Second, we discuss the use of confirmatory factor analysis (CFA) in scale revision only sparingly because more extended treatments of this topic are available elsewhere (see Finch & West, 1997, or Floyd & Wida- man, 1995). Finally, although this point m ay seem obvious, we assume that th e construct being measured is appropriate for a factor analytic representation. That is, we only consider situations in which dimensional latent variables (factors) account fo r indica- tor (item) correlations. We do not consider modeling emergent variables (Bollen & Lennox, 1991; Lykken, McGue, Tellegen, & Bouchard, 1992), latent types (Strube, 1989; Waller & Meehl, 1998), or multifaceted trait concepts (Carver, 1989; Hull, Lehn, & Tedlie, 1991). Overview: Factor Analysis as Applied to Scale Revision If the misapplication of factor methods continues at the present rate, we shall find general disappointment with results because they are usually meaningless as far as psychological research interpretation is concerned. (Thurstone, 1937, p. 73) Th e goal of scale revision is to improve the psychometric properties—and ultimately the validity—of individual-differences measures. Here we use the term validity to imply that a measure (a ) ha s item content and a corresponding factor structure that is representative of and consistent with what is currently known regarding a construct, (b) has a factor structure that is replicable an d generalizable across relevant populations, and (c) has a clearly interpretable (i.e., univocal) an d relatively precise scaling of indi- viduals along one or more common dimensions. It is these three objectives that we have in mind in this article. Note that these objectives are consistent with Loevinger's (1957) notions of sub- stantive and structural validity. Several review articles have recently addressed the application of EFA to personality and psychopathology test data (e.g., Com- rey, 1988; Finch & West, 1997; Goldberg & Digman, 1994). 287

factor analysis and scale revision_reisewallercomrey_2000

Documents